A speech recognition apparatus which determines, based on a speech inputted thereto by a speaker, individual monosyllables intended by the speaker on a one-by-one basis has been widely known. In addition, a speech recognition apparatus which receives multiple syllables as one word has also been widely known. Such speech recognition apparatuses include one which displays multiple candidate speech patterns on a display unit when they are found as a result of the speech recognition described above such that the speaker selects a desired one of the candidates with a speech input (see, e.g., Patent Document 1).
On the other hand, there is a speech recognition apparatus comprising an operating unit such as a key switch for effectively narrowing down multiple candidates as described above to a few.                Patent Document 1: JP-H8-54894 A (FIG. 2)        
However, when an operating unit such as a key switch is disposed on the periphery of a steering wheel or an instrument panel in a speech recognition system as described above, an extra space is needed. This causes the problems of, e.g., an increased number of wires and an increased number of components.